Example of an RGB Image Compositing Technique

Yesterday, we posted a series of images of Hurricane Earl, using various spectral bands from MODIS to produce colorful composites that provided image enhancement to illuminate specific features within the imagery.  Here, as an example, is how the various channels can be combined to produce the final composite.  In this example, data are provided from a nighttime orbit of the Aqua satellite, with a nadir track nearly coincident with the center of the hurricane circulation, providing a clear, high resolution picture of the storm within the MODIS infrared bands.  Due to the nighttime orbit, image composites that make use of the visible, solar reflectance bands are not available, and this example uses the “nighttime microphysics” color combinations that require only infrared brightness temperatures and their differences.  We will start off with a traditional infrared image of the storm using the 11 um channel:

IR Image

Infrared image of Hurricane Earl acquired from Aqua MODIS around 0620 UTC on August 31, 2010

The final RGB composite is the “sum” of color components representing the basic colors that produce a final pixel color on the monitor: red, green, and blue.  In the “nighttime microphysics” enhancement, contributions to the red shades are determined from the difference in the 12-11 micron infrared brightness temperature.  To provide additional enhancement, the 256 available intensities of red color are concentrated within a range from -4 to +2 K, or about 0.02 K of precision assigned to each red shade.  Values less than -4 K are set to zero (black) and values approaching +2 K increase in red shading, with all values greater than +2 set to the maximum red intensity.  In the image below, the “red channel” is shown, with Hurricane Earl prominently displayed.  Therefore, the brightest red shades are associated with deep, convective cloud tops.  Open water and low clouds still appear as shades of red since they contribute infrared brightness temperatures, but with reduced temperature differences.

R of RGB

Contribution of red intensities to the final RGB composite, based upon a 12-11 um channel differencing applied over a range of -4 to 2 K.

Contributions to green intensities are determined from a difference between the 11um infrared and 3.9 um near-infrared brightness temperatures, focusing on a range from 0 to 10 K.  The greatest differences are assigned to the brightest green shadings, and here emphasize the deepest convection associated with the core and northern portions of Hurricane Earl.  Additional convection is present over northern South America.  Lower cloud tops with less convective activity appear in darker green shades, while some of the thinner ice clouds to the southwest of the circulation produce minimal or negative differences and contribute little to the green intensity.

G of RGB

Contribution of green intensities to the final RGB composite, based upon a 11-3.9 um channel differencing applied over a range of 0 to 10 K.

Finally, we have the blue contribution.  Blue intensities are provided by the 11 um channel, limited to a range from 243 to 293 K.  This image is somewhat similar to a basic infrared image that has not been color flipped.  In an image that has not been color flipped, cold brightness temperatures appear black — it is the inverting of the gray scale that provides the traditional “clouds are white” imagery used in operations and on GOES web imagery.  Here, high blue intensities correspond to warm temperatures, and black colors appear at cold temperatures associated with cooler cloud top temperatures.  Low clouds therefore contribute darker blue (low blue intensity), and open water is bright blue (very warm).

B of RGB

Contribution of blue intensities to the final RGB composite, based upon the 11 um channel, without color flip, and applied over a range from 243 to 293 K.

When these colors are combined from each of the R, G, and B panels, the final product is a full range of colors that comprise the “RGB Composite”.  Near the circulation center, the active convection contributes high red intensities and high green intensities, which combine to shades of yellow.  Some of the granulation and stippling appear due to saturation of the green channel contribution and artifacts of reprojecting the original MODIS data from the swath to the final, mapped area.  However, these yellow shades are limited to areas of active convection within the tropical cyclone and also in northern South America.  Many of the higher and thinner ice clouds only contributed in the red channel, but not significantly, so that appear as dark shades of red.  Meanwhile, open waters contributed to all of the RGB components with emphasis on blue shades, leading to a purple appearance.  Low clouds are offset from the remainder of the darker purple background, since they contributed greater amounts to each of the R, G, B components, which tends to lighten the otherwise purple shade.  Bottom line — there are a number of imagery features that can be enhanced through MODIS channel composites, and future GOES-R channel differencing, beyond our current capabilities.  SPoRT will continue to investigate “best practices” for transitioning these concepts to forecast operations.

RGB

Final RGB composite of Hurricane Earl as viewed by Aqua MODIS, based upon the previous three R, G, B contributions.

2 thoughts on “Example of an RGB Image Compositing Technique

  1. Alex — Yep, since Aqua observed Hurricane Earl at near-nadir, that means that we have excellent CloudSat and AMSR-E coverage of the storm, and perhaps CALIPSO. I took a look at some of those this morning and they’re available. I might make a follow on post tomorrow or next week.

    Andrew

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s